123129 research outputs found
Sort by
Essays in the Economics of Crime
Get PDFThis thesis contains three chapters that analyze the economics of crime in Latin America throughan economic lens. The first chapter develops a new longitudinal dataset on the presence of criminal enterprises in Mexican municipalities. I construct this dataset by processing over 21 million
newspaper articles using natural language models to extract (1) whether an article pertains to organized crime, (2) the names of criminal enterprises and locations mentioned in the text, and (3)
which locations (if any) each criminal enterprise operates in. Using this dataset, I document key
stylized facts about the evolution of the criminal landscape in Mexico. I show that the presence of
criminal enterprises within municipalities has become increasingly fragmented over time, with more
groups operating in the same areas and show that many of these groups have splintered from existing
criminal enterprises.
Chapter two examines how targeting the top leaders of criminal enterprises affects market structure and homicide rates, using Mexico as a case study. While executive capture may diminishorganizational capacity, it can also disrupt the balance of power and induce territorial disputes.
Using the dataset developed in Chapter 1, I find that the capture or assassination of a criminal
leader leads to a 54% local increase in the number of criminal enterprises operating in the affected
region, reflecting the entry of competing groups. This market restructuring drives a 32% increase in
the local homicide rate, highlighting the unintended consequences of high-value targeting.
The final chapter develops a method through which to detect unobserved drug smuggling using
transaction level export data. Exploiting a positive supply shock to the Peruvian cocaine market
from coca eradication policy in Colombia, I study the impact of an exogenously induced increase
in coca prices on the value of exports from Peru, as drug traffickers commonly use other goods to
conceal cocaine shipments. Using variation in Colombian coca production to predict coca leaf prices,
I estimate that a $1USD/KG increase in the farm gate price of coca leaves lead to a 248% increase
in the value of exports from Peruvian provinces that are highly suitable to grow coca to countries
identified by intelligence agencies as cocaine transit countries. Additionally, I find evidence that
characterizes drug traffickers as sophisticated agents who reduce the risk of arrest or seizure through
their selection of shipping methods.Ph.D
The Role of Respiratory Muscle Structure and Function on Post-Lung Transplant Outcomes: Integrating Lived Experiences of Respiratory Symptoms – A Multi-Methods Thesis
Get PDFRespiratory muscle limitations and their association with respiratory symptoms, health-related quality of life (HRQL), and physical function have not been well characterized in the post-lung transplant (LTx) population. This thesis compares respiratory muscle function and LTx outcomes between LTx candidates and recipients. Additionally, it evaluates associations of diaphragm function with clinical outcomes post-LTx and explores the impact of respiratory symptoms through patient experiences. LTx recipients demonstrated greater diaphragm function, fewer respiratory symptoms and better HRQL compared to LTx candidates. Greater diaphragm function was significantly associated with increased inspiratory muscle strength and fewer respiratory symptoms. Qualitative findings highlighted four key themes: 1. pre-LTx respiratory decline, 2. post-LTx respiratory recovery, 3. physical recovery trajectories and challenges, and 4. embracing a new life. Persistent respiratory symptoms in some LTx recipients and associations with worse diaphragm function suggest potential for targeted interventions, such as inspiratory muscle training to optimize recovery.M.Sc
Resource Allocation in Vehicular Communications Using Graph Neural Networks
Get PDFThis work presents three graph-based frameworks for cellular traffic prediction and vehicular network resource management, specifically designed to address challenges posed by dynamic network topologies, heterogeneous communication requirements, and mutual interference. Leveraging Graph Attention Networks (GATs) and Graph Transformer Networks (GTNs), the proposed models capture evolving spatiotemporal patterns and support adaptive decision-making in real-world wireless environments. First, we propose a Spatial-Temporal Graph Attention Network (ST-GAT) for cellular traffic forecasting. Unlike traditional Graph Convolutional Networks (GCNs) that assume fixed graph structures, ST-GAT incorporates a dynamic attention mechanism that updates edge weights and connectivity over time, accurately modeling the shifting interactions among cell towers. Evaluations on real-world cellular datasets show significant improvements in prediction metrics, including a reduction in Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE), leading to more reliable traffic forecasting and resource planning. Second, we introduce a GNN-based topology prediction module combined with a dynamic bandwidth allocation algorithm for vehicular networks. This framework anticipates future network states using historical and real-time data, enabling proactive bandwidth reallocation based on evolving Quality of Service (QoS) requirements. Simulation results demonstrate improved throughput, fairness, and latency compared to baseline approaches, ensuring consistent connectivity under high mobility. Finally, we develop a GTN-based interference-aware channel assignment strategy for Device-to-Device (D2D)-enabled vehicular networks that support both Vehicle-to-Infrastructure (V2I) and Vehicle-to-Vehicle (V2V) links. By modeling the network as a graph where edges encode slow-fading channel state information, the GTN effectively captures interference patterns and QoS constraints. The resulting channel assignment reduces inter-link interference, maximizes V2I capacity, and maintains robust V2V communication, significantly outperforming static and heuristic-based baselines. Together, these contributions enable more accurate traffic forecasting, proactive bandwidth control, and interference-aware channel assignment. Our findings support the development of next-generation intelligent transportation systems and scalable cellular infrastructure capable of adapting to dynamic wireless environments.Ph.D
The DAUNTLESS-EG Bus: Small Satellite System Design and Test
No full textSmall satellites in the 100-500 kg range are becoming increasingly attractive for New Space corporations and government institutions alike due to streamlined bus development and economical rideshare launches to Low-Earth orbit. The content in this thesis summarizes the author's work on system-level testing and novel harnessing development on the small satellite StarBurst Multimessenger Pioneer mission. Extending to future high-power missions, preliminary system design is presented to triple the available payload power on this platform. Further detailed design of a power converter is performed in order to enable such a power system. Successful test results of the power converter demonstrate the viability of the hybrid design approach that has been adopted. Several satellite sub-systems are investigated in this research including electrical power systems, command and data handling, communications, thermal, and attitude control. The work presented in this thesis constitutes significant progress towards eventual low-cost and high-performance small-satellite constellations.M.A.S
The Burden of Tuberculosis Disease in Ontario: A Health Technology Assessment Approach to Inform Policy and Resource Allocation
Get PDFDespite being both preventable and curable, tuberculosis (TB) remains a persistent public health concern in Ontario, with nearly 1,000 new cases reported in 2024. The burden of disease falls disproportionately on newcomers to Canada, with 79% of reported cases occurring among individuals born outside the country. While clinical management is well defined, there remains limited understanding of the full economic impact of TB across the health system, for individuals, and at the societal level. This dissertation aimed to address this gap using a multi-method approach to estimate the costs associated with TB disease in Ontario.The first study was a scoping review of model-based economic evaluations in low-incidence settings. It found that most studies focused narrowly on medical interventions, with limited attention to non-medical interventions despite their established importance in TB prevention and control. The second study used linked population-based administrative data and a matched cohort design to estimate TB-attributable health care costs across defined phases of care, generating generalizable cost estimates specific to the Ontario context. In the third study, a cross-sectional survey captured out-of-pocket costs and productivity-related outcomes among people receiving TB treatment at two centres in Toronto, highlighting the financial burden experienced even within a universal health care system. The final study developed a model-based cost-of-illness analysis to estimate the lifetime costs of TB from the health system, patient, and societal perspectives. By capturing costs from multiple perspectives, this work highlights opportunities to tailor interventions that reduce the financial burden for different payers.
Together, these studies provide a comprehensive assessment of the economic burden of TB in Ontario. They demonstrate the value of integrating administrative data, patient-reported outcomes, and decision-analytic modeling to inform evidence-based decision-making. This work contributes to the broader use of health technology assessment to address the economic dimensions of communicable disease control in low-incidence, high-resource settings.Ph.D
Novel Optimization Methods for Temporal and Predictive Clustering
Get PDFClustering, an unsupervised learning method, aims to group unlabeled samples based on similarity, but modern datasets introduce challenges. First, data often extends beyond static features to temporal sequences. Second, clustering may move beyond the geometric similarity between samples in feature space. Traditional clustering methods struggle with these complexities, as they largely assume static, geometrically separable samples. To address this limitation, this thesis introduces several new clustering approaches formulated as Mixed-Integer Linear Programs (MILP) to guarantee global optimization. Specifically, a Temporal Clustering framework addresses time-dependent data and considers temporal dynamism in cluster assignments and definition. A scalable Linear Predictive Clustering formulation groups samples by shared predictive structures in a non-separable feature space. A novel Granger-causal Clustering integrates temporal dynamics with predictive relationships and provides an interoperable definition via Bounded Box constraints. Collectively, these methods advance clustering by incorporating temporal, predictive, and causal structures in a principled optimization framework.M.A.S
Intravenous Iron Prescribing Patterns in Heart Failure: A Quality Review at St. Michael’s Hospital
Get PDFPharmacy residents have the opportunity to complete a research project during their residency training, which provides them with skills on how to conduct and manage a research project. Projects often represent an area of interest and need that has been recognized by the host institution’s pharmacy department. Projects are presented as a poster at an annual CSHP Ontario Branch Residency Research Night, and many eventually go on to be published in a peer-reviewed journal.Background:
Intravenous (IV) iron is a guideline recommended therapy for patients with iron deficiency and heart failure with reduced ejection fraction. While guideline-recommended, this indication is not part of our institution’s prescribing criteria. We wished to investigate the use of IV iron in our inpatient cardiology population and explore reasons for the inpatient use versus the outpatient use.
Objective:
To use a quality lens to evaluate the use of IV iron in patients admitted for heart failure (HF) to the Heart and Vascular Program and at St. Michael’s Hospital (SMH) and to determine barriers to access in the outpatient setting.
Methods:
Interviews were conducted with prescribing stakeholders in the Heart Failure Clinic using a structured set of pre-arranged questions regarding the use of IV iron. Data were collected and generalized into common themes to identify factors, barriers and/or trends.
A retrospective chart review was conducted for patients admitted from November 2023 to November 2024 who were prescribed IV iron. Process (e.g. percentage of patients with iron studies available), outcome (percentage of those meeting SMH prescribing criteria) and balancing (e.g. adverse effects) measures were collected in addition to patient demographics. Compliance rates to institutional IV iron use criteria and to guideline-directed criteria were recorded.
Results:
Common themes regarding outpatient IV iron utilization included lack of resources to accommodate on-site IV administration, poor intra-institutional accessibility to outpatient IV infusions and lack of adequate connections to extra-institutional alternatives to facilitate IV iron infusion administration.
Out of 38 inpatients, 8 (21%) patients had HF with reduced ejection fraction. Process measures included: availability of iron study results which were available for 36 of 38 patients (95%). Among those with documented intended doses (n=30), 24 (80%) patients received their full doses prior to discharge. Outcome measures included those who met institutional IV iron use criteria n=18 (47%) patients in comparison to n=4 (11%) patients who met guideline-directed criteria, with n=36 (65%) of patients receiving 600 mg of IV iron or less. No adverse events relating to IV iron administration were reported in all patients within the study.
Conclusion:
Outpatient access to IV iron was found to be a major barrier to IV iron therapy. Measured metrics demonstrated appropriate assessment and administration. However, the use of IV iron was predominantly in HF patients with preserved ejection fractions with majority of patients receiving 600 mg of elemental iron or less, both findings contrary to available literature. Further studies are warranted to explore quality initiatives to bridge these identified gaps
Experiences of Ableism and Racism Among Racially Minoritized Youth and Young Adults with Disabilities
Get PDFYouth with disabilities face persistent disability-related discrimination (ableism) but research frequently overlooks the experiences of racially minoritized youth. The purpose of our study was to explore the experiences and perceived impact of discrimination among racially minoritized youth and young adults with disabilities. This qualitative study involved a sample of 15 youth and young adults with disabilities (mean age 22 years) identifying as racially minoritized. A descriptive inductive thematic analysis was applied to the interview data. Our findings highlighted the following themes: (1) types of discrimination (i.e., cultural and family-related ableism, racist ableism, and gendered/sexist ableism); (2) perceived impact of discrimination (i.e., social isolation, avoidance of unwelcoming and unsafe situations, impact on physical and mental health, decisions about identity disclosure, and a lack of access to resources and opportunities); and (3) positive coping strategies (i.e., inclusive and safe spaces, self-advocacy, and social and family supports)
ReSCU-Nets: recurrent U-Nets for segmentation of multidimensional microscopy data
Get PDFSegmenting multi-dimensional microscopy data requires high accuracy across many images (e.g.timepoints or Z slices) and is thus a labour-intensive part of biological image processing
pipelines. We present ReSCU-Nets, recurrent convolutional neural networks that use the
segmentation results from the previous frame as a prompt to segment the current frame. We
demonstrate that ReSCU-Nets outperform state-of-the-art image segmentation models, including
nnU-Net and the Segment Anything Model, in different segmentation tasks on time-lapse
microscopy sequences. Using ReSCU-Nets, we investigate the role of gap junctions during
Drosophila embryonic wound healing. We show that pharmacological blocking of gap junctions
slows down wound closure by disrupting cytoskeletal polarity and cell shape changes necessary
to repair the wound. Our results demonstrate that ReSCU-Nets enable the analysis of the
molecular and cellular dynamics of tissue morphogenesis from multidimensional microscopy
data.M.A.S
Characterizing the NLRP6 Inflammasome in the Human Intestinal Epithelium
Get PDFThe intestinal epithelium functions as a barrier and an active participant in innate immune defense. Inflammasomes are cytosolic multiprotein complexes that assemble in response to PAMPs and DAMPs. In mice, the NLRP6 inflammasome is highly expressed in the intestinal epithelium, activating Caspase-1 to cleave pro-inflammatory cytokines and GasderminD (GSDMD), triggering pyroptosis. However, the mechanism and activating ligands of human NLRP6 remain unclear due to the lack of physiologically relevant models. In this study, we developed a human intestinal organoid model to study NLRP6 signaling. Seeding 3D organoids into 2D monolayers reproducibly induced NLRP6 expression. Using lentiviral shRNA, we established stable knockdowns of NLRP6 and GSDMD. Our results suggest that human NLRP6 responds to intracellular flagellin, identifying it as a potential bacterial sensor. This organoid-based system offers a powerful tool to study human inflammasome pathways and could reveal therapeutic targets for diseases involving inflammasome dysregulation and intestinal barrier dysfunction.M.Sc